mlsanityData to 14 Oct 2026

DN-Hypo-Pipeline: An AI-Driven Workflow for Generating Hypotheses using Large Language Models and Scientific Explanations

arXiv 2606.08532 ↗First posted 7 Jun 2026cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Modern artificial intelligence excels at prediction but cannot explain. From large language models to AI-for-science systems, today's machines answer what by recombining patterns already present in the human literature, yet they cannot reason out why a phenomenon must arise from underlying principles even though explanation, not prediction, lies at the heart of scientific discovery. Here we ask whether the structure of scientific explanation can be operationalized to guide how a machine generates hypotheses. We introduce DN-Hypo-Pipeline, a hypothesis-generation framework that adopts a layered, explanation-theoretic scaffold: Hempel's deductive-nomological (DN) model supplies the output form and deductive validity of a hypothesis, Salmon's causal-process account supplies an organizing constraint on where to search for the governing laws, and Armstrong's view of laws as relations between universals supplies the bridge from a phenomenon's constituent processes to the laws that may be associated with it. Rather than searching the space of what has been written, the framework searches the space of what principles govern a phenomenon: given an explanandum, it abstracts the universals instantiated in the phenomenon's formation process, retrieves the laws relating those universals, and deductively reconstructs a new, testable explanation. Evaluated in data-science modeling and judged by both LLMs and human experts, hypotheses generated through this principled reasoning significantly outperform those from direct prompting. Crucially, we translated the two highest-scoring hypotheses into novel algorithms one that reduces the Transformer's theoretical complexity with only minimal performance loss, and another that achieves competitive accuracy with substantially fewer parameters.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 35 of its 69 references that are in the map. First posted on arXiv 7 Jun 2026.

Builds on

35 of 69 references.

References · 35

  1. 2017Attention Is All You Needcited by 60,361
  2. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  3. 2013Efficient Estimation of Word Representations in Vector Spacecited by 4,375
  4. 2013Distributed Representations of Words and Phrases and their Compositionalitycited by 3,474
  5. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,211
  6. 2020Transformers are RNNs: Fast Autoregressive Transformers with Linear Attentioncited by 2,022
  7. 2016Enriching Word Vectors with Subword Informationcited by 1,583
  8. 2024A Survey on LLM-as-a-Judgecited by 1,423
  9. 2024From Generation to Judgment: Opportunities and Challenges of LLM-as-a-judgecited by 494
  10. 2025Accelerating scientific discovery with Co-Scientistcited by 401
  11. 2024Can LLMs Generate Novel Research Ideas? A Large-Scale Human Study with 100+ NLP Researcherscited by 363
  12. 2018Adaptive Input Representations for Neural Language Modelingcited by 226
23 more
  1. 2019Transformer Dissection: A Unified Understanding of Transformer's Attention via the Lens of Kernelcited by 213
  2. 2024ResearchAgent: Iterative Research Idea Generation over Scientific Literature with Large Language Modelscited by 210
  3. 2016Theory-guided Data Science: A New Paradigm for Scientific Discovery from Datacited by 171
  4. 2026Towards End-to-End Automation of AI Researchcited by 159
  5. 2023Large Language Models for Automated Open-domain Scientific Hypotheses Discoverycited by 100
  6. 2025Towards Scientific Intelligence: A Survey of LLM-based Scientific Agentscited by 98
  7. 2024MOOSE-Chem: Large Language Models for Rediscovering Unseen Chemistry Scientific Hypothesescited by 72
  8. 2025ResearchBench: Benchmarking LLMs in Scientific Discovery via Inspiration-Based Task Decompositioncited by 66
  9. 2024ContiFormer: Continuous-Time Transformer for Irregular Time Series Modelingcited by 63
  10. 2017Hash Embeddings for Efficient Word Representationscited by 55
  11. 2017Compressing Word Embeddings via Deep Compositional Code Learningcited by 35
  12. 2024Large Language Models as Biomedical Hypothesis Generators: A Comprehensive Evaluationcited by 34
  13. 2024ChemReasoner: Heuristic Search over a Large Language Model's Knowledge Space using Quantum-Chemical Feedbackcited by 33
  14. 2017Mimicking Word Embeddings using Subword RNNscited by 30
  15. 2025PiFlow: Principle-Aware Scientific Discovery with Multi-Agent Collaborationcited by 28
  16. 2024Retrieval Augmented Fact Verification by Synthesizing Contrastive Argumentscited by 25
  17. 2025Sparks of Science: Hypothesis Generation Using Structured Paper Datacited by 25
  18. 2024Improving Scientific Hypothesis Generation with Knowledge Grounded Large Language Modelscited by 22
  19. 2023Dynamic Retrieval Augmented Generation of Ontologies using Artificial Intelligence (DRAGON-AI)cited by 14
  20. 2023Harnessing the Power of Adversarial Prompting and Large Language Models for Robust Hypothesis Generation in Astronomycited by 11
  21. 2025HypER: Literature-grounded Hypothesis Generation and Distillation with Provenancecited by 10
  22. 2024LLM4GRN: Discovering Causal Gene Regulatory Networks with LLMs -- Evaluation through Synthetic Data Generationcited by 7
  23. 2026Principle-Evolvable Scientific Discovery via Uncertainty Minimizationcited by 7

Built on by

Not cited yet.